Wednesday, January 15, 2014

Book-A-Day 2014 #15: How to Lie With Statistics by Darrell Huff

I'll avoid my usual joking around and elliptical introductions and just be blunt: this is one of the short list of books that every adult should read. How to Lie with Statistics is a classic -- originally published in 1954 -- and the examples and (occasionally) the buried assumptions of Huff make that very clear, but it's still one of the clearest and best examinations of what numbers can do and what people can make numbers do.

Huff, interestingly, was not a professor of math, but a jobbing writer, who fell into a profitable line of "how to" articles and books, of which Lie With Statistics is the most famous. (The others seems to be more serious, and to actually teach their readers how to do the things they promise.) He also lived to the age of 87, and only died in mid-2001; sixty-year-old works of art are often startlingly close to the modern day.

How to Lie With Statistics does, in a sly way, exactly what it promises to do: explain how elementary statistics works and how various statistical concepts -- mean vs. median vs. mode "averages," biased samples, probable error -- can be misused or deliberately skewed. He also cover the abuses of graphs and charts, and the importance of choosing the right frame of reference and sample in getting exactly the numbers you want.

Some of Huff's assumptions are quaint, and a few may even be offensive to some audiences. But his reasoning and examples are as good today as they were in 1954, and there's no other book -- particularly not one so short, pointed, and perfectly titled -- that can do what How to Lie With Statistics does.

If it does nothing else, it'll make you start asking that core question of journalism: who is telling me this, and what does he want to convince me of? Again, this is a book everyone should read and take to heart: it contains an essential toolkit for living in the modern world of advertising, politics, and marketing.

Book-A-Day 2014 Introduction and Index

No comments:

Post a Comment